Computational science demands extreme performance because the running time of an application often determines the size of the experiment that a scientist can reasonably compute. Unfortunately, traditional compiler technology is ill-equipped to harness the full potential of today's computing platforms, forcing scientists to spend time manually tuning their application's performance. Although improving compiler technology should alleviate this problem, two challenges obstruct this goal: hardware platforms are rapidly changing and application software is difficult to statically model and predict. To address these problems, this thesis presents two techniques that aim to improve a compiler's adaptability: automatic resource characterization and selective, dynamic optimization. Resource characterization empirically measures a system's performance-critical characteristics, which can be provided to a parameterized compiler that specializes programs accordingly. Measuring these characteristics is important, because a system's physical characteristics do not always match its observed characteristics. Consequently, resource characterization provides an empirical performance model of a system's actual behavior, which is better suited for guiding compiler optimizations than a purely theoretical model. This thesis presents techniques for determining a system's data cache and TLB capacity, line size, and associativity, as well as instruction-cache capacity. Even with a perfect architectural-model, compilers will still often generate suboptimal code because of the difficulty in statically analyzing and predicting a program's behavior. This thesis presents two techniques that enable selective, dynamic-optimization for cases in which static compilation fails to deliver adequate performance. First, intermediate-representation (IR) annotation generates a fully-optimized native binary tagged with a higher-level compiler representation of itself. The native binary benefits from static optimization and code generation, but the IR annotation allows targeted and aggressive dynamic-optimization. Second, adaptive code-selection allows a program to empirically tune its performance throughout execution by automatically identifying and favoring the best performing variant of a routine. This technique can be used for dynamically choosing between different static-compilation strategies; or, it can be used with IR annotation for performing dynamic, feedback-directed optimization.

Anthropogenic climate change is approaching predicted tipping points and there is an urgent need to de-carbonize energy systems on a global scale. Generation technologies that do not emit greenhouse gas need to be rapidly deployed, and energy grids need to be updated to accommodate an intermittent fluctuating supply. Rapidly advancing battery technology, cost reduction of solar and wind power and other emerging generation technologies are making the needed changes technically and economically feasible. Extracting energy from fast-flowing tidal currents using turbines akin to those used in wind farms, offers a reliable and predictable source of GHG free energy. The tidal power industry has established the technical feasibility of tidal turbines, and is presently up-scaling deployments from single isolated units to large tidal farms containing many turbines. However there remains significant economic uncertainty in financing such projects, partially due to uncertainty in predicting the long-term energy yield. Since energy yield is used in calculating the project revenue, it is of critical importance. Predicting yield for a prospective farm has not received sufficient attention in the tidal power literature. this task has been the primary motivation for this thesis work, which focuses on establishing and validating simulation-based procedures to predict flows through large tidal farms with many turbines, including the back effects of the turbines. This is a challenging problem because large tidal farms may alter tidal flows on large scales, and the slow-moving wake downstream of each rotor influences the inflow to other rotors, influencing their performance and loading. Additionally, tidal flow variation on diurnal and monthly timescales requires long-duration analysis to obtain meaningful statistics that can be used for forecasting. This thesis presents a hybrid simulation method that uses 2D coastal flow simulations to predict tidal flows over long durations, including the influence of turbines, combined with higher-resolution 3D simulations to predict how wakes and local bathymetry influence the power of each turbine in a tidal farm. The two simulation types are coupled using a method of bins to reduce the computational cost within reasonable limits. The method can be used to compute detailed 3D flow fields, power and loading on each turbine in the farm, energy yield and the impact of the farm on tidal amplitude and phase. The method is demonstrated to be computationally tractable with modest high-performance computing resources and therefore are of immediate value for informing turbine placement, comparing turbine farm-layout cases and forecasting yield, and may be implemented in future automated layout optimization algorithms. / Graduate